Dual Adaptive Model Predictive Control with Disturbances

Model-based control requires good accuracy of model parameters to achieve high performance. Controller design under parametric uncertainties is therefore a challenging topic in control system engineering. One well known control design for uncertain systems is adaptive control. An adaptive controller has two tasks: process regulation and parameter learning. Dual control explores the trade-off between the two seemingly conflicting tasks. The control structure in an adaptive system consists of a model-based controller and a recursive update rule for parameter estimation. In this paper, the adaptive control framework consists of model predictive control for system regulation and recursive least squares for parameter estimation. The requirement for persistent excitation is shown to be necessary for systems with high-dimensional parameter space. Then, a dual formulation is proposed in an attempt to generate excitation signals while maintaining control performance in the adaptive control scheme. The algorithm is implemented and tested on a simulated SISO system.